Which transcript? A genetic report can look reassuringly precise: a gene, a string of letters and numbers, and a predicted change in a protein. Then somebody asks which transcript was used. A different transcript can change the numbering. More awkwardly, it can change whether the relevant stretch of DNA is treated as a protein-coding exon at all. Today I went looking at the history behind MANE Select, the shared reference transcripts that now appear throughout clinical genetics. The epilepsy genes offer some particularly good reasons to care. An exon can be present in the sequencing data and still remain outside the diagnostic conversation. I find this an uncomfortable achievement for something called an annotation.
Figure. Two alternative exon choices in SCN8A. MANE Select currently pairs NM_001330260.2 with ENST00000627620.5 and includes exon 5A; MANE Plus Clinical pairs NM_014191.4 with ENST00000354534.11 and includes exon 5N. These are alternative transcripts, not two exons included together. The schematic shows only the alternative exon and generic flanking exons, not the full transcript, genomic distances, or expression levels. Transcript designations were checked against ClinGen on November 11, 2026; the isoform assignments are described by Vanoye et al., 2024. The historical failure to annotate exon 5A should not be mistaken for its absence from current MANE Select. See NCBI MANE guidance for the distinction between Select and Plus Clinical. A transcript designation does not establish variant pathogenicity. Original schematic by Fleck; no patient-level or unpublished data.
Several versions. A gene can produce different RNA transcripts through alternative splicing and different transcript starts or ends. For a protein-coding gene, those transcripts may encode different protein isoforms. This creates two practical problems. The same genomic variant can acquire different transcript-based descriptions, making reports difficult to compare. Alternatively, an exon used by one transcript may be absent from another, so a variant’s annotated consequence changes with the reference. These problems helped motivate the MANE collaboration. The acronym means Matched Annotation from NCBI and EMBL-EBI. “Select” identifies a jointly chosen representative transcript, with RefSeq and Ensembl/GENCODE records that have identical sequence and splice structure and match the GRCh38 reference assembly. It is a useful place to begin reading a gene.
A shared reference. NCBI and EMBL-EBI announced MANE in October 2018, building on their Consensus Coding Sequence collaboration. The ambition extended beyond agreeing on a protein: the paired transcripts would have matching sequences and exon structures, including their untranslated ends. Selection combined biological evidence with expert curation rather than simply picking the longest transcript. The first release, in December 2018, covered 53% of human protein-coding genes. By June 2021, coverage had reached 97%, as described in the 2022 Nature paper. Those dates matter. MANE emerged from years of reconciling two major annotation systems, while both continued to retain alternative transcripts. The point was to make the default reproducible without pretending that every gene had become simple.
Room for another exon. One representative transcript cannot contain two mutually exclusive exons at the same position. MANE therefore also provides MANE Plus Clinical: additional transcripts needed to describe publicly reported pathogenic or likely pathogenic variants that Select alone cannot adequately represent. The 2020 release already included 43 of these additional transcripts. This is particularly relevant to ion channels, where alternative exons can encode functionally important pieces of the protein. Select and Plus Clinical provide a common foundation for reporting. They do not constitute an exhaustive catalogue of every biologically relevant transcript, and the presence of an exon in either set does not establish that a particular variant within it causes disease.
The missing fifth exon. The clearest epilepsy example comes from SCN8A. In a 2017 report, published in print in 2018, the Epilepsy Genetics Initiative reanalyzed exomes from 54 previously undiagnosed trios. Two individuals had de novo missense variants in alternative exon 5A; a third was identified in an additional clinical laboratory cohort. One of the variants had previously been reported as an intronic variant of uncertain significance. Exon 5A had only recently entered the Consensus Coding Sequence annotation. The sequence was therefore capable of carrying a diagnosis before the annotation used in the original interpretation adequately represented it. BTIC returned to these findings when discussing exome reanalysis in 2019. This is a concrete reason to revisit an older negative result.
The history has moved on. It would be misleading to turn that story into a claim that today’s MANE Select misses SCN8A exon 5A. The current ClinGen entry lists NM_001330260.2 as MANE Select and NM_014191.4 as MANE Plus Clinical. These represent the 5A and 5N isoforms, respectively, often called adult and neonatal. A 2024 study also showed that isoform and cellular context can influence the functional effects measured for SCN8A variants. The historical diagnostic lesson and the present annotation are therefore compatible: the alternative exon problem was real, and today’s paired transcript choices help address it. A functional experiment still needs to specify which channel isoform was tested.
An older calcium-channel lesson. CACNA1A supplies another genuine alternative-exon example, with an important phenotypic qualification. In 2008, Graves and colleagues reported three truncating variants in alternative exon 37A associated with episodic ataxia type 2. The alternate exon 37B was unaffected. This demonstrated that preserving one isoform did not compensate sufficiently for disruption of the other. The current MANE choices include NM_001127222.2 as Select and NM_001127221.2 as Plus Clinical. A concrete variant represented on the latter is NM_001127221.2:c.5569C>T, p.Arg1857Ter, recorded in ClinVar. CACNA1A is central to epilepsy genetics, but this historical exon-specific evidence concerns episodic ataxia. Keeping that distinction is part of interpreting the example correctly.
Beyond the channel isoform. SCN1A adds a different mechanism. In 2018, Carvill and colleagues described seven likely pathogenic variants outside annotated coding exons and provided evidence that five promoted inclusion of a poison exon. Inclusion of exon 20N introduces an unproductive transcript and reduces full-length SCN1A output. This is different from choosing between two ordinary protein-coding channel isoforms. The disease mechanism involves altered RNA processing in a region that a coding-exon-only analysis may not adequately assess. This broadens the lesson from the SCN8A story: alongside the choice of a protein-coding transcript, diagnosis may depend on understanding how the cell regulates productive RNA.
Reading the small print. A useful habit is to retain the genomic reference and the transcript accession together, including their versions. The HGVS checklist explicitly calls for an accession with its version number. When comparing an old paper with a current report, check that the descriptions map to the same genomic change rather than assuming that similar protein names settle the question. For a plausible alternative-exon finding, the next questions concern the relevant transcript, disease mechanism, tissue and developmental expression, and the laboratory’s coverage and annotation. A MANE designation helps orient that review. It does not replace variant classification, establish assay coverage, or turn every rare alternative-exon variant into an explanation.
What you need to know. MANE Select has made genetic reporting easier to compare by providing matched reference transcripts across two major annotation systems. MANE Plus Clinical preserves additional clinically important transcript choices when one representative is insufficient, as the project guidance explains. For epilepsy genetics, the historical SCN8A exon 5A diagnoses show why annotation can determine whether a finding reaches interpretation. CACNA1A supplies an older alternative-exon lesson from episodic ataxia, while SCN1A extends the discussion to poison-exon regulation. My practical takeaway is to start with the shared reference and remain curious about what it leaves out. The accession number in the small print can change the meaning of the entire report.
